Multi-threaded

[spencer-lunarg]

I guess curious if there is any reason we need to only run one test at at time.

We do something similar in VVL generating our SPIR-V from GLSL files (an LLM will be better to tell you how to use python though)

        with ThreadPoolExecutor() as executor:
            futures = {executor.submit(process_shader, shader, gpu_shaders_dir, glslang, spirv_opt, args.targetenv): shader for shader in shaders_to_compile}

            for future in futures:
                shader = futures[future]
                try:
                    result = future.result() # This will raise exceptions if any occurred inside process_shader
                    compiled_shader_data.append(result)
                except Exception:
                    compilation_failed = True

The main thing if doing this is if multiple tests fail, then it should make sure to only print one at a time